From a pile of voice notes to a report you can act on
A fictional café put two QR codes on its premises and asked customers how their visit went. Below: five of the voice messages exactly as they were said, then the report the café gets from all 150 of them. Nothing here is a real business or a real customer.
1 · What comes in
Raw voice messages, as people said them
Five of the 150 messages. No script, no multiple choice: people talk the way they talk, mix Hindi and English, and pause to think. Messages that are not in English carry a machine translation underneath.
- Message 1HI · HindiCounter QRVoice message length: 0:34
भैया, लाइन में बीस मिनट खड़ा रहा… बस एक ही आदमी बिल बना रहा था। चाय अच्छी थी, लेकिन इतना इंतज़ार, अब दोबारा सोचूँगा।
Machine translation: Brother, I stood in the line for twenty minutes… only one person was doing the billing. The tea was good, but this much waiting, I'll think twice now.
AI reads it as: Wait time at the counter, negative
- Message 2EN · EnglishTable tent QRVoice message length: 0:22
Uh, so the paneer sandwich was really good, like properly fresh, and the, um, the filter coffee is honestly the best I've had around here. Yeah. That's it.
AI reads it as: Food and drinks, positive
- Message 3HI · HindiCounter QRVoice message length: 0:27
वो नीली शर्ट वाला लड़का है न, उसने मेरी बेटी के लिए बिना कहे बिना चीनी की हॉट चॉकलेट बना दी। बहुत अच्छा लगा।
Machine translation: The boy in the blue shirt, you know, he made a sugar-free hot chocolate for my daughter without being asked. It felt really nice.
AI reads it as: Staff friendliness, positive
- Message 4HI · HindiTable tent QRVoice message length: 0:19
कोल्ड कॉफ़ी एक सौ अस्सी की है, साइज़ देखो तो थोड़ी महँगी लगती है। बाकी सब ठीक है।
Machine translation: Cold coffee is a hundred and eighty rupees; for the size it feels a bit expensive. Everything else is fine.
AI reads it as: Prices and value, negative
- Message 5EN · EnglishTable tent QRVoice message length: 0:25
The table was sticky when we sat down and, uh, nobody wiped it until I asked. Also the fan above table four is really loud. Otherwise nice place though.
AI reads it as: Cleanliness and seating, negative
2 · What comes out
The same voices, as a report
Every report opens with how the study was made, then the findings. Each theme links back to the message its quote came from, so nothing is taken on trust.
Café visit feedback, 1–7 Oct 2026
Sample — fictional businessHow this study was made
- Sample source
- Open QR codes: anyone who scanned a code at the counter or on a table tent. Not an invited list.
- Field dates
- 1–7 Oct 2026
- Scanned, started, valid
- 214 scans, 176 started, 150 valid, 26 rejected
- Why messages were rejected
- Silent or too short 14, duplicate 8, no consent 4
- Language and channel mix
- Hindi 78, English 72. All 150 answered as voice messages.
- Completion
- 85% of those who started left a valid message
- Known bias
- Only people who chose to scan are heard. Counter codes sit by the billing queue, so counter answers lean towards people who waited.
- Quota control and margin of error
- No quotas were applied, so no margin of error is shown. Findings are directional only unless quota controls were applied and met.
Headline numbers
- 150
- valid voice messages
- 44%
- positive overall
- 26%
- mention wait time at the counter first
- 85%
- of those who started finished
What people talked about
Each message is counted once, under its main theme.
Wait time at the counter
39 messages, 26%
3 positive, 6 neutral, 30 negative
“only one person was doing the billing”
From message 1 Food and drinks
36 messages, 24%
26 positive, 4 neutral, 6 negative
“the paneer sandwich was really good, like properly fresh”
From message 2 Staff friendliness
30 messages, 20%
24 positive, 3 neutral, 3 negative
“made a sugar-free hot chocolate for my daughter without being asked”
From message 3 Prices and value
24 messages, 16%
8 positive, 8 neutral, 8 negative
“for the size it feels a bit expensive”
From message 4 Cleanliness and seating
21 messages, 14%
5 positive, 4 neutral, 12 negative
“nobody wiped it until I asked”
From message 5
How people felt overall
- Positive 44% (66)
- Neutral 17% (25)
- Negative 39% (59)
Complaints worth acting on
Ranked by how many negative messages each theme holds. Actions are suggestions; you decide.
Long queue at the billing counter
30 negative messages, 51% of all negative. 23 at the counter, 7 on table tents.
Suggested action: Open a second billing point at busy hours; try QR pre-ordering for take-away.
Tables left sticky; one loud fan
12 negative messages, 20% of all negative. 4 at the counter, 8 on table tents.
Suggested action: Add a table wipe-down round every half hour; have the fan above table four checked.
Cold coffee feels small for the price
8 negative messages, 14% of all negative. 4 at the counter, 4 on table tents.
Suggested action: Review the cold coffee portion or price against the menu card.
By QR code location
| Counter | Table tent | |
|---|---|---|
| Messages | 84 (56%) | 66 (44%) |
| Positive | 42% | 47% |
| Negative | 43% | 35% |
| Most talked about | Wait time at the counter (29) | Food and drinks (20) |
Want a report like this for your own customers?
Tell us who you would ask and what you would like to learn. We reply within one working day.
The café, its customers and every number on this page are invented for illustration. This is not a result from a real study.